1) To take standard implicit solvent models, combined with atomistic biomolecular models and modern sampling techniques, and attempt to directly evaluate the absolute free energy of ligand binding in a range of protein-ligand systems. 2) To develop statistical methods to identify of which, if any, of various sampling techniques are able to truly able to yield precise free energies of binding complexes for a range of relevant systems. 3) To use these results to identify ways in which the implicit solvent and atomistic biomolecular models can be improved to better fit the binding energies and structures for a wide variety of ligand protein complexes. This research has important consequences for rational drug design, and consequently all of medicine. If the predictive ability of biomolecular simulations can be significantly improved, and the efficiency of these methods increased, it will result in significantly lowered total costs of drug development.